Time-Series Prediction Task

A Time-Series Prediction Task is a sequence prediction task whose input is a (possibly unlabeled) temporal dataset.

• Context:
  • It can range from being a Heuristic Temporal Prediction Task to being a Data-Driven Temporal Prediction Task (such as a supervised temporal prediction task).
  • It can range from being a Temporal Point Prediction Task to being a Temporal Series Prediction Task (such as timeseries pattern classification).
  • It can range from being a Time-Series Categorical Prediction Task to being a Time-Series Ranking Task to being a Time-Series Numeric Prediction Task.
  • It can be solved by a Temporal Prediction System (that implements a temporal prediction algorithm).
  • It can be supported by a Time-Series Smoothing Task.
• Example(s):
  • a Forecasting Task, such as temperature prediction and future demand prediction task.
  • an Anomaly Detection Task, such as a fraud detection.
  • predict what the thrown object will do next.
  • a Time-to-Event Prediction Task.
  • …
• Counter-Example(s):
  • a Timeseries Clustering Task.
  • an I.I.D. Prediction Task.
  • a Text String Prediction Task.
• See: Temporal Data Analysis, Timeseries Labeling.

---

---